A linear regression model is said to be good when the r-squared value tends to 0, 1,-1 or 0.5
Answers
Answer:
Explanation:
R-squared can never surpass the 0–1 distance. Intuitively, r-squared represents how much the independent variables describe the percentage of the variability in the target and dependent variable. It has to be between 0 percent -100 percent or between 0-1 because it's a percentage.
0 R-squared shows the "rock bottom" as the independent variables did not explain any of the variability. And a 100% R-square means that the model has perfectly captured the variance.
R² ranging from -1 to + 1 is not an indicator on the basis of which a model is good or otherwise. Yet R² proximity to either -1 or + 1 suggests a possible overfit.
Typically an 80–90% r-square is all right. Occasionally, ppl uses another statistics called modified r-squared which is an improvement over r-squared.